Combining color and shape descriptors for 3D model retrieval

Abstract

Nowadays many three dimensional models feature color information together with the shape description. However current content-based retrieval schemes for 3D models are based on shape information only and ignore color clues. The significance of shape versus color clues for 3D model retrieval is instead a fundamental issue still almost unexplored at this time. A possible approach is to extend shape-based 3D model retrieval methods of proven effectiveness in order to include color. This work follows such rationale and introduces an extended version of the spin-image descriptor that can account also for color data. The comparison of color descriptors is performed using a novel scheme that allows to recognize as similar also objects with different colors but distributed in the same way over the shape. Shape and color similarity are finally combined together by an algorithm based on fuzzy logic. Experimental results show how the joint use of color and shape data allows to obtain better results than each of the two types of information alone. Comparisons with state-of-the-art content-based retrieval methods for 3D models also show how the proposed scheme outperforms standard solutions on object classes with meaningful color information.

Recent Trends, Applications, and Perspectives in 3D Shape Similarity Assessment

“…Finally, in case of additional photometric information (texture), structures is usually captured by embedding shapes in both the Euclidean space and in a colour‐based one (such as the RGB or the CIELab colour space) or possibly in a larger one somehow combining the above two. Examples of related methods include those based on colour‐aware generalizations of purely geometric approaches such as heat kernel signatures [KBBK12], geodesic distance [BCGS13] and spin‐images [PZC13]; other methods take inspiration from techniques initially conceived for 2D images, as in the case of [TSDS11, ZBH12] with the SIFT algorithm [Low04].…”

Recognition of feature curves on 3D shapes using an algebraic approach to Hough transforms

“…Examples of these descriptions are the Local Binary Patterns (LPB) [52], the Scale Invariant Feature Transform (SIFT) [47], the Histogram of Oriented Gradients (HOG) [17] and the Spin Images [35]. The generalization of these descriptors to 3D textured models has been explored in several works, such as the VIP description [71], the meshHOG [73] and the Textured Spin-Images [53]. Further examples are the colour-CHLAC features computed on 3D voxel data proposed in [37]; the sampling method [46] used to select points in regions of either geometry-high variation or colour-high variation, and to define a signature based on feature vectors computed at these points; the CSHOT descriptor [66], meant to solve point-to-point correspondences coding geometry-and colour-based local invariant descriptors of feature points.…”

New Compact 3-Dimensional Shape Descriptor for a Depth Camera in Indoor Environments

“…Spin Image was built by constructing a 2D histogram of the distances and heights of the neighboring points. Several methods to improve the discriminative power of Spin Image have been proposed [ 17 , 19 , 20 ]. Pasqualotto et al [ 20 ] proposed to combine color and shape Spin Images to compare two 3D models where similarities of two types of Spin Images are aggregated by fuzzy logic.…”